If you are looking for DNA binding proteins then double stranded is the usual choice. Oligos or fragments can be 25-70 bp or even larger. 32-P is used to end label the oligos (10-50 ng) with kinase. The probe design is limited by the sequences you wish to examine. Use small amounts of oligo, after end labeling, we generally use about 2000-4000 cpm per binding reaction. Gels should be 6% neutral acrylamide with 0.5X TBE running buffers. This is old school but still works.

I have only one point to add to this perfect "old school" protocol. We use large chambers and cool them during the run.